CHANGES IN COCOA PROTEINS
Purr, A., Springer, R., Morcinek, H., Z. Lebensm.-Unters.-Forsch. 112, 40 (1960). Quesnel, V. C., Jugmohunsingh, K., J . Sci. Food Agric. 21, 537 (1970). Reineccius, G. A., Keeney, P. G., Weissberger, W., J . Agric. Food Chem. 20, 202 (1972). Roelofsen, P. A., Adv. Food Res., 225-296 (1958). Wolf, G., Nahrung 2, 948 (1958). Zak, D. L., Keeney, P. G., J . Agric. Food Chem. 22,800 (1974).
Zak, D. L., Keeney, P. G., J . Agric. Food Chem., following paper in this issue (1976). Received for review July 9, 1975. Accepted December 22, 1975. Published as paper no. 4886 on June 16,1975 in the Journal series of the Pennsylvania Agricultural Experiment Station. Supported in part by funds provided by the Chocolate Manufacturers Association of the U S A .
Changes in Cocoa Proteins during Ripening of Fruit, Fermentation, and Further Processing of Cocoa Beans Dennis L. Zakl and Philip G. Keeney*
During ripening of cocoa fruit the protein content of seeds decreased 25 ?& , but consistent qualitative trends were not apparent. Most of the protein of unfermented beans could be solubilized and recovered. However, at the conclusion of fermentation only about one-third of the protein could be extracted from beans. The extractable protein fractions became progressively less pure during fermentation and a t the end were only 40% protein. Fewer, more diffuse protein bands were evident on disc gels as fermentation advanced. Most of these changes are believed to involve protein interaction with polyphenols. Somewhat similar trends were noted during roasting and conching.
A recent study (Zak and Keeney, 1976) revealed protein differences between the two fundamental types of cacao, white Criollo and purple, pigmented Forastero. Compared to the latter, less than half as much protein could be solubilized and extracted from Criollo cocoa beans and, after a series of steps to remove contaminants from the extract, the protein content of the final dehydrated material was only about one-half that obtained from Forastero varieties. Moreover, electrophoresis of Criollo material yielded fewer protein bands and distinct differences were found when classified according to solubility characteristics. These differences relate to both genetically derived protein variations and the effects of post harvest events, especially phenolic tanning reactions. This presentation is a complement to the above study. Using almost identical procedures, information was collected concerning protein changes during ripening of cocoa fruit, post harvest fermentation of cocoa beans, and key processing steps in the chocolate factory, namely, roasting and conching. Cocoa fruit (pods) require 4-5 months to grow to full size following pollination (Seeschaaf, 1971). At this time the beans contained therein have also reached near maximum development. The pods are then left on trees to ripen for about a month before being harvested. During ripening the mucilaginous pulp surrounding the beans undergoes changes critical to a successful fermentation, primarily an increase in fermentable carbohydrate (Rohan, 1963). After harvesting, the seeds and mucilaginous pulp from opened pods are fermented several days before the beans Department of Food Science, The Pennsylvania State University, Borland Laboratory, University Park, Pennsylvania 16802. 'Present address: M & M/MARS, Division Mars, Inc., Hackettstown, New Jersey 07840.
are dehydrated. Fermentation results in the formation of precursors of essential aroma compounds eventually generated by roasting in the chocolate factory. Fermentation also brings about a suppression of astringency and bitterness. Proteolysis, enzymatic and nonenzymatic browning, and tanning are important sequences in the development of chocolate flavor and other characterizing properties identified with chocolate. Birch (1941) reported a significant loss of protein nitrogen during fermentation, only a portion of which would be accountable to proteolysis or diffusion through the testa. As suggested by DeWitt (19571, a more imporant contributor would be protein insolubilization caused by oxidation and tanning reactions involving polyphenols and polyphenol oxidase. Forsyth et al. (1958) showed that protein-polyphenol interactions do occur during fermentation to reduce protein solubility. Polyphenols combine with proteins in a manner analogous to tannins converting animal hides to leather (Forsyth and Quesnel, 1963). This involves reversible complexing through hydrogen bonds, and irreversible oxidation of polyphenols to quinones followed by covalent condensation of quinones with reactive groups of amino acids, peptides, and proteins (Loomis, 1969). The latter most probably is the cause of the limited solubility of protein in fermented cocoa beans. A detrimental result of tanning is the poor biological value reported for cocoa beans (Lanteaume et al., 1972). Tanning does, however, have effects on chocolate which are advantageous. The burnt feather taste of roasted protein is depressed as is the astringency associated with polyphenols (Forsyth and Quesnel, 1963). Thus, from the standpoint of flavor these interactions are desirable to a certain extent. The processing of cocoa beans in the chocolate factory involves roasting followed by grinding of the cotyledon or nib portion to yield chocolate liquor, which may be separated into cocoa butter and cocoa fractions. Chocolate J. Agric. Food Chem., Vol. 24, No. 3, 1976
483
ZAK, KEENEY
is obtained by mixing, refining, and conching a blend of chocolate liquor, cocoa hutter, and sugar, with or without milk solids. The imDortance of roasting bas been referred to meviously. donching is a process in which molten chocolate is under continuous movement or agitation for a period ranging from a few hours to several days, during which unwanted aromas are volatilized, and oxidation along with other poorly understood chemical reactions occur. The net effect of conching is a mellowing of the flavor of chocolate. Aeration and maximum surface exposure appear to be especially important. MATERIALS AND METHODS
Source of Samples. Samples for a study of protein changes during ripening of cocoa fruit were collected at the Cocoa Research Center. Itabuna, Bahia. Brazil. The pods were opened and testa and adhering pulp material were peeled from the cotyledons, which were then dried at 60 "C under reduced pressure. Cocoa heans used in the fermentation study were supplied by V. C. Quesnel, Cocoa Research Department, University of the West Indies, Trinidad. Samples were taken daily during a 6-day, sweat box fermentation. A comparison between unroasted and roasted cocoa beans was made using Bahia beans roasted 30 min at 150 'C in a laboratory, forced air oven. Conching was evaluated using commercial sweet chocolate conched 24 h at a paste temperature of 65 "C in a Petsholtz PVS 100 vertical conche. Protein Recovery a n d Analysis. Procedures reported by Zak and Keeney (1976) were followed to recover protein fractions from cocoa beans and sweet chocolate. This included extraction nf far with ethyl ether and polyphenols with methanol. Defatted cocoa mass t0.E-1.0 g) with most nnl.inhnnnl;r lm oinr;nl llyylllyl
yyAJrLL"LavLLL
romnsroA
lllllvIly
sx,oo DvtrrrtDA &rh u ~ v L ~ ~ u 1-1111 u u
rror;r
ub..uAb
acid-urea-detergent solution, and the extract was filtered and concentrated after adsorption of residual phenols by a nonsoluble form of polyvinylpyrrolidone. A column of " . L . J . . .
TTTO,,
rxynituex L r l ' U
.....
.... J I. W a b "bell L U
.._L.._ _. ^I_: , L _ A . . . Jeyaraw _ yruLeIII UUl"
_^I
11u11-
protein components, and the protein fraction was dialyzed and freeze dehydrated to yield a protein "prep". Quantitative protein data were obtained for fractions off the Sephadex column by the method of Lowry et al. (1951) and by amino acid analyses for protein preps. Proteins in protein preps were separated by disc gel electrophoresis, and the preps were separated into solubility classes by successive treatment of undissolved material with water, 10% NaC1,70% aqueous ethanol, and finally 0.2% NaOH. Again, methods followed were those descri bed by Zak and Keeney (1976). RESU LTS AND DISCUSSION
Rip from poas sampieo mrougnour ripening are Iuuscrareu in Table I. Protein content of seeds decreased progressively between 135 and 1 130 days. Likewise, the amount of protein that could be solubilized decreased. Since the data were collected on a pooled weight basis rather than for individual seeds, it c()uld not be determined if protein was being consumed dur ing ripening or if other components in moisture-free defalked heam were increasing. The latter seems more likely. Schmieder (1976) found a slight increase in starch conttmt during ripening in the same series of bean samples. During ripening t hle purity of the protein prep material increased between 135 and 150 days and then decreased sharply in the later sitages. Reasons for this are not clearly evident. Interpretation is complicated by the questionable 484
J. Agric. F O O ~Chem , Vol 24, NO 3, 1976
-
Table I. Protein Changes in Seeds of Cjoeoa rrun " during Ripening Days aifter pollination ~
135 1'13
150
160
Defatted cocoa bean mass Total protein,o % , 23.6 2!1.6 21.4 17.6 Extractable protein,b% 14.2 1.3.2 12.5 11.5 Protein prep material Purity,= % protein 57.7 76.0 81.5 54.7 Solubility classification 42.4 57.0 57.0 44.2 Albumin, % Globulin, % 7.8 9.3 9.3 8.8 Prolamine, % 16.5 15.7 19.4 18.6 Glutelin, 9% 32.2 16.6 13.0 28.5 a From amino acid analysis of hydrolyzed defatted cocoa bean mass. Protein off Sephadex by the method of Lowry .n1Ymoil . . et. al. (1951). . . From amino acid analysis of ,mter,a1.
Disc el electrophoresis of protein preps beans at different stages of ripening. Post pollination: (1) 160 days; ( 2 ) 150 days; (3)143 days; (4)135 days. ~
~ I I lecoyere(l lrOm ~~~~~
~~~
~
..-
. .-,:,I:+ -C+h.. .. l.... "'l""LbJ " 1 b u r "-"r
,.h+":--,l fI" ,1 .- thn ?9K-An.r ""L-lr" b1.r ' U V - u n J
onmnln
n:-mo
~au~y~ DL . IzI L, C
it may have been an artifact related to the recovery of seed material from immature fruit. Whereas seeds from older fruit were obtained intact, those from the 135-daysample were fragile and shattered into spongy sections during removal of the testa. Disruption of cellular integrity and mixing of protein and polyphenols to form insoluble complexes is suggested. This could account for the low purity of the 135-day sample. Unfortunately, logistical problems vis-a-vis our laboratory and Brazil forced postponement of the confirmatory studies. Solubility trends for protein preps during ripening show alhumins and glutelins varying inversely to each other (Table I). The least and most ripe samples, 135 and 160 days, could be paired together according to similarity in solubility characteristics, as could the 143- and 150-day protein preps. Disc gels (Figure 1)do not show contrasting differences, although bands for preps of 143- and 150-day samples appear more diffuse and more complex than is the case for the 135-day and fully ripe, 160-day cocoa samples. This may reflect a concentration effect. The same amount of material was applied to all gels, but since samples of intermediate ripeness were more pure (Table I) a greater amount of protein was being applied to the gels for separation. Acid hydrolysis of protein preps followed hy amino acid analysis did not reveal differences which could be related to stage of ripening. Amino acid patterns were similar to those reported by Zak and Keeney (1976). Fermentation. Changes relating to protein parameters recorded for a series of samples collected during a fer-
CHANGES IN COCOA PROTEINS
Table 11. Protein Changes in Cocoa Beans during Fermentation Fermentation days Defatted cocoa bean mass Total protein: % Extractable protein,b % Protein prep material Purity,' % protein Solubility classification Albumin, %
0
1
2
3
4
5
6
17.6 17.1
17.2 15.4
20.4 12.8
20.7 10.4
21.2 10.0
20.8 8.0
20.1 7.4
72.7
70.0
63.8
49.1
50.1
35.0
41.1
56.1 24.2 3.8 15.9
72.5 13.7 3.1 10.7
41.3 47.3 21.0 28.2 Prolamine, % 4.6 12.8 Glutelin. % 19.8 24.8 a From amino acid analysis of hydrolyzed cocoa bean mass. e From amino acid analysis of hydrolyzed prep material. Globulin, %
mentation of Trinidad cocoa hems are presented in Table 11. The finding that total protein content changed only slightly during fermentation is in agreement with Maravalhas (1972). Except for a slight loss in the sweatings during the later stages of fermentation, degradation produds are retained within the bean complex. The s m a l l increase in protein occurring early in the fermentation probably reflects microbial synthesis of amino acids and proteins in the pulp. The percentage composition of the amino acid fraction did not change appreciahly during fermentation. The laree and almost linear decrease in extractable protein during fermentation (Table 11)is contrary to the results of Niepage (1961) who reported increases in extractable protein during fermentation. It is believed his data are in error because the procedures employed would favor tanning and contamination of the extract. Our results conform with the sequences suggested by DeWitt (1957). Heat and acid generated by microorganisms canse bean death, after which polyphenols and proteins are released from their respective cells and tanning occurs. As tanning progresses, proteins hecome increasingly more insoluble, During the early stages, proteolytic enzymes hydrolyze proteins to more water-soluble components, smaller peptides, and amino acids (DeWitt, 1957; Roelofsen, 1958; Rohan and Stewart, 1967). This would account for a portion of the decrease observed, since these smaller components would not hecome part of the extractable protein fraction (Birch, 1941; Wolf, 1958). The reduction in protein prep purity indicates progressively more contamination of the proteins with polyphenols. The contention of Birch (1941) that 50% of the protein is hydrolyzed or made an insoluble complex appears conservative. The first 1-3 days of fermentation are especially critical and coincide with bean death. Location in the fermentation heap influences when individual beans die (Lopez, 1974); thus, mass viability is quite heterogeneous in the early stages. Although cocoa protein contains a full complement of amino acids, aspartic and glutamic acids together account for 25% of the total in fresh seeds (Zak and Keeney, 1976). Pronounced changes in these acids during fermentation were noted. Aspartic acid decreased progressively from 12.4 to 7.9% of total amino acids between 0 and 6 days. During this period glutamic acid increased from 15.4 to 26.2%. Generally, most other amino acids decreased, but less so than aspartic acid. The reason for the increase in glutamic acid while most other amino acids were decreasing is not clearly nnderstood. One possibility is that cocoa protein contains a suhstantial quantity of glutamine. Since amides of dicarboxylic acids show less reactivity with polyphenols (Mason and Peterson, 1965), glutamate accumulation
70.7 15.0 3.8 10.5
Protein off Sephaaex oy mernoo 01 Lowry et al. (1951).
1
Figure 2. Disc gel electrophoresis of protein preps trom cocoa beans at various stages of fermentation. Fermentation time left to right (1)through (7): 0,1,2, 3, 4, 5, and 6 days, respectively. Table 111. Protein Changes Caused by Roasting Cocoa Beans Roasted
Raw
Defatted cocoa bean mass
Total protein,a % 18.4 17.1 Extractable protein? % 4.2 3.5 Protein prep material Solubility classification Albumin, % 62.9 79.5 Globulin, % 12.1 3.6 Prolamine, % 14.2 6.8 Glutelin, % 10.9 9.7 From amino acid analysis of hydrolyzed cocoa bean ~~~~~~
h n ~ ~ l - d C" . _: l . . l .~ .. ~
L
......
.P.
^I
^I
might he expected during fermentation. Proportioning the proteins of the prep material into solubility classes showed increases in the albumin fraction with accompanying decreases in globulins, prolamines, and glutelins in the early stages of fermentation. Albumin increases could reflect hydrolysis of water-insoluble proteins by proteolytic enzymes. Relative proportions did not change greatly after about the third day which could coincide with bean death and reduced proteolytic activity. Disc gel electrophoresis of protein preps revealed subtle changes in proteins during fermentation (Figure 2). Bands from unfermented cocoa above region A and between A and B eventually disappeared as fermentation progressed. The most striking change occurred near 2-3 days when beans die. Below region B hands diffused and became fuzzy after 1-2 days. Similar results, noted for the proteins of heated milk, have been credited to interactions of protein components (Manning, 1969). For cocoa this may J. Agric. Food Chem.,
VoI. 24, NO.3.
1976
485
ZAK, KEENEY
Table IV. Protein Changes Caused by Conching
of Sweet Chocolate
Conching time, h
n -
Figure 3. Disc gel electrophoresisof protein preps from unroasted (1)and roasted ( 2 ) Bahia cocoa beans. reflect interaction of proteins and/or polyphenol complexing. Roasting a n d Conching. As revealed in Table I11 roasting caused a slight decrease in the protein content of cocoa beans. In this study total protein was calculated from amino acid data for hydrolyzed samples and would include the contribution of free amino acids in beans. Involvement of free amino acids in nonenzymatic browning reactions during roasting is well documented (Bailey et al., 1962; Pinto and Chichester, 1966) and could account for most of the recorded decrease in protein content. Undoubtedly some protein amino acids, particularly lysine, are also destroyed. Reduction of extractable protein during roasting is attributed to Maillard browning and further polyphenolic tanning. While enzymatic activity would not be expected, air roasting would provide ample oxygen for autocatalytic oxidation of polyphenols and continued tanning of proteins. Solubility studies of protein preps yielded trends similar to those observed during fermentation. The albumin content of the protein fraction of the protein prep increased from 63 to 80% balanced by decreases in globulins, prolamines, and glutelins. Disc gels (Figure 3) of Bahia heans showed the same protein bands, but they were less intense than in raw cocoa. The total protein content of sweet chocolate was not changed appreciably by conching (Table IV). Values obtained through amino acid analyses, 4.3 to 4.7% of defatted sweet chocolate mass, are similar to the protein content estimated by assuming 28% liquor in the paste (information from supplier). If liquor is 10% protein (Watt and Merrill, 19631,defatted sweet chocolatewould be 4.4% protein by calculation. As indicated previously, the amount of extractable protein is reduced greatly by fermentation and roasting. Table IV shows that extractable protein values, already low due to these processes, are further reduced by conching. After 24 h of conching less than 10% of the protein in chocolate was recovered in the extractable protein fraction. A similar effect was noted for the purity of the protein prep. Only 5% of the prep material was protein after 24 h. Almost all of this protein fell into the water-soluhle, albumin class, which again follows trends occurring during fermentation and roasting. Disc gels showed no distinct hands, only areas of limited fluorescence. SUMMARY As shown through the data reported herein, cocoa protein undergoes progressive alteration during fermen-
486
J. Agric. Food Chem., VoI. 24, NO. 3, 1976
__
-"
?A
15
Defatted sweet chocolate mass Total proteinP % 4.4 4.7 4.3 Extractable protein,b % 0.7 0.5 0.4 Protein prep material Purity,e % protein 10.9 9.8 5.0 Solubility classification 98.2 98.4 100.0 Albumin, % Globulin, % 1.8 1.6 0.0 Prolamine, % 0.0 0.0 0.0 Glutelin, % 0.0 0.0 0.0 From amino acid analysis of hydrolyzed, defatted chocolate. Protein off Sephadex by method of Lowry P -...:~~. -.:~, .~..,...:. .P,...>..., .... > ~I -1
I.,%*
